Objective HgCdTe detectors have an important application in hyperspectral detection, their performance directly determines the system's detection sensitivity and measurement accuracy. However, when they are used across a wide dynamic range, pronounced response nonlinearity often occurs. This nonlinearity not only limits the detectors' effective operating range but also significantly compromises spectral inversion accuracy in the applications of Fourier transform-based hyperspectral detection. Therefore, it is important to study the response nonlinearity of HgCdTe detectors, clarify its mechanisms, and explore effective improvement strategies, which have both significant theoretical significance and engineering value.
Methods A systematic experimental and theoretical investigation was conducted to study the influence of equivalent series resistance on response nonlinearity. Photovoltaic detectors with different electrode contact areas were fabricated. Their I-V characteristics were tested, and their response nonlinearity was quantitatively measured using the dual-aperture method at different bias voltages. Based on experimental research, a two-dimensional physical model of the device was established by integrating TCAD with circuit models. The model comprehensively accounts for multiple factors such as internal light absorption within the device, carrier generation and recombination, as well as thermal effects. This model was used to simulate the effects of different electrode contact areas, bias voltages, p-region doping concentrations, and p-region carrier mobilities on the response nonlinearity. Finally, based on the results, directions for process optimization efforts were proposed.
Results and Discussions The fabricated HgCdTe detectors with n+-on-p structure remain essentially linear at a photon flux density of 1×1021 s−1·m−2. At 2×1021 s−1·m−2, they exhibit response nonlinearity (Fig.3). With R_\mathrms/R_0^' increases, the response nonlinearity increases and can be fitted by a quadratic relation within a certain range. When R_\mathrms/R_0^' < 6\% , the nonlinearity exhibits reduced sensitivity to series resistance. Theoretical analysis based on the equivalent circuit model of photovoltaic detectors indicates that when signal current is output, the presence of series resistance causes the p-n junction operating point to shift into forward bias. This results in a corresponding decrease in the device's photocurrent, making it one of the primary factors affecting detector linearity. Moreover, applying reverse bias to the device effectively reduces the space charge shielding effect by increasing the external electric field, thereby decreasing the response nonlinearity. Devices with lower series resistance exhibit a more significant reduction in nonlinearity since the voltage drop allocated at the depletion region is greater (Fig.7). A two-dimensional model was established based on experimental results to simulate the variation of nonlinearity with series resistance (Fig.9). The simulation results are consistent with the results of the experimental test, which validates the reliability of the model. When the same reverse bias is applied to the devices, a lower series resistance leads to a greater increase in the electric field across the depletion region. This also results in a larger reduction in nonlinearity (Fig.10 and Fig.11). Then how changes in the series resistance of the neutral region, caused by variations in the p-region doping concentration and carrier mobility, affect the response nonlinearity was analyzed. Increasing the p-region doping concentration significantly reduces the nonlinearity because it lowers the series resistance (Fig.12). However, increasing the carrier mobility increases the nonlinearity due to a rise in the signal current. Nonetheless, at the same output current level, a device with higher mobility exhibits lower nonlinearity (Fig.13).
Conclusions The response nonlinearity of HgCdTe photovoltaic detectors was analyzed from theoretical, experimental, and simulation. The results indicate that the electrode contact area, doping concentration and material thickness can effect nonlinearity. The fabrication of HgCdTe detectors with high linearity in a large dynamic range can be achieved by optimizing electrode preparation prosses, adopting sub-pixel structures, appropriately reducing substrate thickness, etc. Applying an appropriate reverse bias can also eliminate nonlinearity. In practical fabrication, the contact resistance can be reduced through the optimization of the metal-semiconductor contact and the electrode preparation process. Additionally, a sub-pixel structure can be adopted, where a large photosensitive area is designed as numerous small photosensitive units connected in parallel. This design decreases the equivalent length of the neutral region, thereby lowering its resistance. Furthermore, nonlinearity can also be reduced by improving the passivation process to increase the zero-bias resistance and by raising the doping concentration in the p-region.